I remember about 40 years ago, the CICS development team were very pleased with their work optimising the CICS dispatcher code - the key code was written in assembler, and polished till they could get no more benefits from tuning it. Then came the next generation hardware, and all their polishing was of no benefit. They found it was best to structure their code and let the PLS compiler optimise it - which was generally better than they could. Colin
On Fri, 22 Aug 2025 at 08:47, Paul Edwards < [email protected]> wrote: > On Thu, 21 Aug 2025 21:22:13 -0700, Leonard D Woren < > [email protected]> wrote: > > >> Writing and maintaining code requires actual work by a skilled engineer, > >> so there is a limit to how big it can be. > > > >What if you're using one of those bloatware languages that I > >previously dissed? How big is the executable for "hello world" in > >various languages? Did you know that saving an M$Word file with > >contents of only "hello world" is 24K? (At least it was in the > >mid-90s when I tried it.) > > While it's true that I was thinking of assembler and C when I > wrote my original comment, in particular 400,000 lines of C > code produces a 3 MB executable typically - that is indeed > making use of an optimizer - but hand-written assembler can > obviously achieve the same thing. > > While I haven't looked at the assembler produced by other > languages, I would assume that they are the same as > unoptimized C, or not much worse. I was once involved in > a (large) project written in Ada and from memory they > managed to get a 16 MiB executable that exceeded some > system limitations. The fact that the vendor hadn't seen > that before suggests it was rare. > > So yeah - even if my guess of DB2 being a few MB of > actual executable is wrong - I don't think it's going to be > wrong by a factor of 100 or anything like that. > > And even a factor of 100 would be a pretty small proportion > of a 4 GiB address space, which is what the original > question was about. ie why does anyone need to put their > code above 2/4 GiB? Short answer - they don't. But I want > to do it anyway, for reasons other than "need". > > BTW, the z390 suggestion elsewhere in this thread has led > to me being able to get my C90-compliant programs built > to run on MVS to z/OS on a PC using gccmvs plus z390. > ie I am no longer dependent on having an MVS environment > to target MVS. That's still 32-bit though (aka 31-bit, aka > 24-bit). I'm planning on eliminating use of official MVS > assembler macros and replacing them with direct calls to > SVCs etc, so that I only need a smattering of assembler > that doesn't need macro facilities, so I can use as370 and > eliminate the need for z390 too. > > And then implement another technique to override the SVC > so that I can run under z/PDOS-generic (and mfemul as > the emulator) so that I can run (certain) perfectly valid > z/OS executables on an MSDOS-like system. > > BFN. Paul. > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
